View as
Switch to see what matters for your role
The coverage gap map
349 enumerated provisions × 4 axes (G, S, P, C) + care_cost lens = 1,312 cells. The finding: the field has measured almost none of them.
Field coverage — the headline
1,301 of 1,312
cells have no measurement in any known benchmark
Blind cells by axis
Each row is one axis. The filled portion is the cells with any field measurement; the empty portion is what the instrument renders as blind.
Five reasons a cell is blind
Each blind cell falls into one of these categories. The distribution shows where the gaps are most concentrated.
no_benchmark847No benchmark
No instrument exists in the field for this cell.
wrong_granularity312Wrong granularity
Field benchmarks exist but at category, not provision, granularity.
speaker_only89Speaker only
Benchmark asks 'would the model answer compliantly?' — not 'would it act compliantly?'
bare_model_only42Bare model only
Benchmark scores a base model, not a deployed agent.
judgement_based11Judgement-based
Benchmark uses a model judge; not deterministic; rejected by Law 1.
The 11 field-evidenced cells
Every covered cell below cites its source benchmark in that source's own units, with licence. Each is named so a reviewer can find it.
| Provision | Axis | Source | Licence | GSPC? |
|---|---|---|---|---|
| Art 5(1)(c) — social scoringEU AI Act | S | DefBench care battery (regulation-derived refusal, 5 items) | internal · signed | measured |
| Art 5(1)(f) — emotion inference at workEU AI Act | S | DefBench care battery (regulation-derived refusal, 3 items) | internal · signed | measured |
| Art 14 — human oversightEU AI Act | G | Bench-2-CoP (human-oversight coverage at category granularity) | research-use | not measured |
| Art 5 prohibited practices (care_cost)EU AI Act | care_cost | GSPC production sweep | internal | measured |
| Sch 1 Part 1 — special category conditionsUK DPA 2018 | G | ICO guidance | OGL v3.0 | not measured |
Internal only — never the headline: 3 of 1312 cells have been measured by GSPC. Reported here only to disambiguate; the product is the map of the field's obligation-space blind spots, not our coverage of them.
Why this matters to you
Different audiences see different value in the gap map. Here's how each stakeholder uses it.
For Investors & VCs
The 99.2% blind spot is the moat. Every compliance-AI startup claims coverage; this map proves none have it. We are the only ones mapping the actual obligation space.
- → 1,301 unmeasured cells = addressable market
- → 99.2% blind = competitive vacuum
- → Deterministic predicates, never a model judge = defensible
- → Published refutations = trust signal
For Regulators
A read-only instrument you can audit. Every cell cites its source in the source's own units. Every anchor is timestamped. No claim is made that evidence does not support.
- → 6 live anchor registries (OGL, EU reuse, CC BY)
- → Tamper-evident chain (sha256 → Ed25519)
- → No LLM in the verdict (Law 1)
- → Self-scoring disclosure (we measure ourselves)
For IP & Legal Counsel
Published measurement cards can be Ed25519-signed and linked to an exact root. That proves specific bytes and signer; it does not make every run tamper-evident or establish legal acceptance.
- → Chain integrity verified (DR-0032, DR-0033)
- → 9 refutations published (honest corrections)
- → Sample sizes labelled (n<20 = lower bound)
- → Methodology open (predicates are deterministic)
🔒 IP defensibility
The novelty is not the rule set — the rule set is public. The novelty is the deterministic harness + signed chain + published refutation discipline. That combination cannot be reverse-engineered from a competitor demo.
📋 Regulatory admissibility
Each "[MEASURED]" tag carries a chain link. A regulator can request the hash, verify the chain independently, and reach the same number without trusting our server. This is the BAR for admissibility under EU AI Act Art 12.
💼 Diligence-ready
The 9 refutations are the diligence asset. A founder who publishes what killed their own bets is a founder who cannot surprise an LP with a hidden failure. This is the moat that ships with the team.